From patchwork Fri Nov 22 07:59:39 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Masahiro Yamada X-Patchwork-Id: 179952 Delivered-To: patch@linaro.org Received: by 2002:a92:38d5:0:0:0:0:0 with SMTP id g82csp4739338ilf; Fri, 22 Nov 2019 00:00:55 -0800 (PST) X-Google-Smtp-Source: APXvYqxZpmk+lR8r1T17IpA1j1HxKG7y5KSWulrxrKMaOiJHoIXXKtwk6ci4jphLQL4vJ71OSFLE X-Received: by 2002:a17:906:27cc:: with SMTP id k12mr20937011ejc.181.1574409655251; Fri, 22 Nov 2019 00:00:55 -0800 (PST) ARC-Seal: i=1; a=rsa-sha256; t=1574409655; cv=none; d=google.com; s=arc-20160816; b=G4AzX5AWISVQ3JxYn9vmk0zpVBwjr/BGci8AJNx+xXYB/sU8H2waXSa4+5pArAyXpM rwvzdVY6pyWiA/x9+p/w0sh5epsjWhupPFzp/ZUl2T1T+68lAtxAZntq12kg0YRzRjxI 4VVxlX8qjSW42dTTjQyPcoeV1nsvQPqj3dxHa4qUO2vFf8ONZqAut76MYqx4zTxorOzJ Due/i6tkAkz9Dye+Xz2RBVqcSXh3m9S6+tN3o8rRjEwgCPvdO8r7ys1BY2ACEyPgQOui FvU0QRU9DQqGx0T7W0Niy7rkIzzANqJyshO2YJFl+J4H82uMkGjh1WHiHgV+injbA3K7 +mkA== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=list-id:precedence:sender:message-id:date:subject:cc:to:from :dkim-signature:dkim-filter; bh=fPMOgceq6VcAzfubQqqmhDdamDoMjnT4yAQ7KXJflwg=; b=brmc/I6HRK/lj9AvAyrBVwJU4rY4ZeoD2ZdrtaMQn0gxH4hq0vCwqnC9GYb4nXCgVj 5mFgaAwAQuY0geHiRC+nejIyp2E5IosP4aXeJMu+cWNxHODrV0p6d4JV6C5Yt+mswf6O 1n1kGbWFD6rdJMopx9zGeB0RNxknjLb1HupOn+5K23GbYjkUJNgAYJudkAdHaSJZtOCS uCTde5K5pVnGYQ4u09mVUp3GW2DfAwzfVM9iFhBKzBchBmkle2+FWC96Zo1YlgY7l4nS EGVC2SiZij9/JC3GFBPRHIUsfRuJHZY/VHDTvEQ2HIp7EO4lvZ+xm9yu3H3iq8Ui/mb8 TSUQ== ARC-Authentication-Results: i=1; mx.google.com; dkim=pass header.i=@nifty.com header.s=dec2015msa header.b=RRHl36rt; spf=pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org Return-Path: Received: from vger.kernel.org (vger.kernel.org. [209.132.180.67]) by mx.google.com with ESMTP id d14si3812050ejc.72.2019.11.22.00.00.54; Fri, 22 Nov 2019 00:00:55 -0800 (PST) Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67; Authentication-Results: mx.google.com; dkim=pass header.i=@nifty.com header.s=dec2015msa header.b=RRHl36rt; spf=pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1726836AbfKVIAt (ORCPT + 26 others); Fri, 22 Nov 2019 03:00:49 -0500 Received: from conuserg-09.nifty.com ([210.131.2.76]:17390 "EHLO conuserg-09.nifty.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726018AbfKVIAs (ORCPT ); Fri, 22 Nov 2019 03:00:48 -0500 Received: from localhost.localdomain (p14092-ipngnfx01kyoto.kyoto.ocn.ne.jp [153.142.97.92]) (authenticated) by conuserg-09.nifty.com with ESMTP id xAM7xjvb022344; Fri, 22 Nov 2019 16:59:45 +0900 DKIM-Filter: OpenDKIM Filter v2.10.3 conuserg-09.nifty.com xAM7xjvb022344 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=nifty.com; s=dec2015msa; t=1574409586; bh=fPMOgceq6VcAzfubQqqmhDdamDoMjnT4yAQ7KXJflwg=; h=From:To:Cc:Subject:Date:From; b=RRHl36rtyPe/KbIFZZnO6a4IophfcABjLMSTeZa05LkhEDnoDZD4e67Z5EpfEHIq+ qVbvjJ0j+Cyo7rrzuCVi9VYWKhHZBikU+OHiWCCJFaL4XF4oYBaZU8YSr8XDIqCkRx h5Jo+8QaJDrU2eYFHMLvm/+wStJmvvP3Bhlm7FQYaiFeuzQXBRpv/ig6ZjWp/4cJDH WxRkWgjXiJ1lkNL/4u7Rv6MIQEJeaUaxFwvTrrNjC6vrGnLltbsDBglLZmbYcztd+w 05cyDFDw7g8C32JWkkJHibGbi9gyb4vjS/YUW9dGiajKiDwEXktLl4sdLoHwI/0SOQ JwNS+f0lMHWaQ== X-Nifty-SrcIP: [153.142.97.92] From: Masahiro Yamada To: linux-kbuild@vger.kernel.org Cc: Arnd Bergmann , Stephen Rothwell , Paul Gortmaker , Masahiro Yamada , Greg Kroah-Hartman , Lucas De Marchi , linux-kernel@vger.kernel.org Subject: [PATCH] kbuild: check MODULE_* macros in non-modular code Date: Fri, 22 Nov 2019 16:59:39 +0900 Message-Id: <20191122075939.14481-1-yamada.masahiro@socionext.com> X-Mailer: git-send-email 2.17.1 Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Paul Gortmaker sent a lot of patches to remove orphan modular code. You can see his contributions by: $ git log --grep='make.* explicitly non-modular' To help this work, this commit adds simple shell-script to detect MODULE_ tags used in non-modular code. It displays suspicious use of MODULE_LICENSE, MODULE_AUTHOR, MODULE_DESCRIPTION, etc. I was not sure about module_param() or MODULE_PARM_DESC(). A lot of non-modular code uses module_param() to prefix the kernel parameter with the file name it resides in. If we changed module_param() to core_param(), the interface would be broken. MODULE_PARM_DESC() in non-modular code could be turned into comments or something, but I am not sure. I did not check them. I built x86_64_defconfig of v5.4-rc8, and this script detected the following: notice: asymmetric_keys: MODULE macros found in non-modular code notice: binfmt_elf: MODULE macros found in non-modular code notice: bsg: MODULE macros found in non-modular code notice: compat_binfmt_elf: MODULE macros found in non-modular code notice: component: MODULE macros found in non-modular code notice: debugfs: MODULE macros found in non-modular code notice: drm_mipi_dsi: MODULE macros found in non-modular code notice: freq_table: MODULE macros found in non-modular code notice: glob: MODULE macros found in non-modular code notice: intel_pstate: MODULE macros found in non-modular code notice: n_null: MODULE macros found in non-modular code notice: nvmem_core: MODULE macros found in non-modular code notice: power_supply: MODULE macros found in non-modular code notice: thermal_sys: MODULE macros found in non-modular code notice: tracefs: MODULE macros found in non-modular code notice: vgacon: MODULE macros found in non-modular code To fix above, check MODULE_LICENSE(), MODULE_AUTHOR(), etc. Please check #include , THIS_MODULE, too. I confirmed they are all valid. Maybe the 'debugfs' is unclear because there are tons of debugfs stuff in the source tree. It is talking about MODULE_ALIAS_FS() in fs/debugfs/inode.c because fs/debugfs/debugfs.o never becomes a module. [How to fix the warnings] Let's take 'asymmetric_keys' as an example. (1) grep Makefiles to find the relevant code $ git grep -A2 asymmetric_keys -- '*/Makefile' '*/Kbuild' crypto/Makefile:obj-$(CONFIG_ASYMMETRIC_KEY_TYPE) += asymmetric_keys/ crypto/Makefile-obj-$(CONFIG_CRYPTO_HASH_INFO) += hash_info.o crypto/Makefile-crypto_simd-y := simd.o -- crypto/asymmetric_keys/Makefile:obj-$(CONFIG_ASYMMETRIC_KEY_TYPE) += asymmetric_keys.o crypto/asymmetric_keys/Makefile- crypto/asymmetric_keys/Makefile:asymmetric_keys-y := \ crypto/asymmetric_keys/Makefile- asymmetric_type.o \ crypto/asymmetric_keys/Makefile- restrict.o \ Then, you notice it is associated with CONFIG_ASYMMETRIC_KEY_TYPE and is a composite object that consists of asymmetric_type.o, restrict.o, ... (2) Confirm the CONFIG is boolean $ git grep -A2 'config ASYMMETRIC_KEY_TYPE' -- '*/Kconfig*' crypto/asymmetric_keys/Kconfig:menuconfig ASYMMETRIC_KEY_TYPE crypto/asymmetric_keys/Kconfig- bool "Asymmetric (public-key cryptographic) key type" crypto/asymmetric_keys/Kconfig- depends on KEYS Now you are sure it never get compiled as a module since ASYMMETRIC_KEY_TYPE is a bool type option. (3) Grep the source file(s) $ grep '^MODULE' crypto/asymmetric_keys/asymmetric_type.c MODULE_LICENSE("GPL"); Remove the orphan MODULE tags. You may also need to do some additional works such as: - replace module_*_driver with builtin_*_driver - replace with - remove module_exit code - move credit in MODULE_AUTHOR() to the top of the file Please see Paul's commits. Signed-off-by: Masahiro Yamada --- scripts/modules-check.sh | 54 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 54 insertions(+) -- 2.17.1 diff --git a/scripts/modules-check.sh b/scripts/modules-check.sh index f51f446707b8..5f1d98d4ee30 100755 --- a/scripts/modules-check.sh +++ b/scripts/modules-check.sh @@ -13,4 +13,58 @@ check_same_name_modules() done } +# Check MODULE_ macros in non-modular code +check_orphan_module_macros() +{ + # modules.builtin.modinfo is created while linking vmlinux. + # It may not exist when you do 'make modules'. + if [ ! -r modules.builtin.modinfo ]; then + return + fi + + # modules.builtin lists *real* built-in modules, i.e. controlled by + # tristate CONFIG options, but currently built with =y. + # + # modules.builtin.modinfo is the list of MODULE_ macros compiled + # into vmlinux. + # + # By diff'ing them, users of bogus MODULE_* macros will show up. + + # Kbuild replaces ',' and '-' in file names with '_' for use in C. + real_builtin_modules=$(sed -e 's:.*/::' -e 's/\.ko$//' -e 's/,/_/g' \ + -e 's/-/_/g' modules.builtin | sort | uniq) + + show_hint= + + # Exclude '.paramtype=' and '.param=' to skip checking module_param() + # and MODULE_PARM_DESC(). + module_macro_users=$(tr '\0' '\n' < modules.builtin.modinfo | \ + sed -e '/\.parmtype=/d' -e '/\.parm=/d' | \ + sed -n 's/\..*//p' | sort | uniq) + + for m in $module_macro_users + do + warn=1 + + for n in $real_builtin_modules + do + if [ "$m" = "$n" ]; then + warn= + break + fi + done + + if [ -n "$warn" ]; then + echo "notice: $m: MODULE macros found in non-modular code" + show_hint=1 + fi + done + + if [ -n "$show_hint" ]; then + echo " To fix above, check MODULE_LICENSE(), MODULE_AUTHOR(), etc." + echo " Please check #include , THIS_MODULE, too." + fi +} + check_same_name_modules +check_orphan_module_macros